Let me endeavour to make this intelligible to the intelligent but unscientific reader. Suppose the Pyramid of Cheops shown for the first time to a giant whose eye was on such a scale that he could just discern it as a separate object. He might make all sorts of ingenious conjectures as to its nature, but if microscopes had been invented in Giant-land and he looked through one, he would find that it was built up, layer by layer, on a regular plan and in determinate lines and angles, by molecules, or what seemed to him almost infinitely small masses, of squared stone. For pyramid write crystal, and we may see by the human sense, aided by human instruments and human reason, a similar structure built up in the same way by minute particles. Or again, divide and subdivide our iron filings until we reach the limit of possible mechanical division discernible by the microscope; each one remains essentially a bar of iron, as capable of being magnetised, and showing the same qualities and behaviour under chemical tests as the original bar of iron from which the filings were taken. This carries us a long way down towards the infinitely small, for mechanical division and microscopic visibility can be carried down to magnitudes which are of the order of 1/100000th of an inch. But this is only the first step; to understand our molecules we must ascertain whether they are infinitely divisible, and whether they are continuous, expanding by being spread out thinner and thinner like gold-beater’s skin: or are they separate bodies with intervals between them, like little planets forming one solar system and revolving in space by fixed laws. Ancient science guessed at the former solution and embodied it in the maxim ‘that nature abhors a vacuum’: modern science proves the latter. In the first place bodies combine only in fixed proportions, which is a necessary consequence if they consist of definite indivisible particles, but inconceivable if the substance of each is indefinitely divisible. Thus water is formed in one way and one only: by uniting one volume or molecule of oxygen with two of hydrogen, and any excess of one or the other is left out and remains uncombined. But if the molecules could be divided into halves, quarters, and so on indefinitely, there can be no reason why their union should take place always in this one proportion and this only.